1. No category

IEEE TRANSACTIONS ON ELECTRON DEVICES, VOL. 61, NO. 11, NOVEMBER 2014
3661
A Steep-Slope Tunnel FET Based SAR
Analog-to-Digital Converter
Moon Seok Kim, Student Member, IEEE, Huichu Liu, Student Member, IEEE, Xueqing Li, Member, IEEE,
Suman Datta, Fellow, IEEE, and Vijaykrishnan Narayanan, Fellow, IEEE
Abstract— This paper explores the energy efficiency advantage
of a 6-bit III-V heterojunction tunnel field-effect transistor
(HTFET) based successive-approximation register (SAR) analogto-digital converter (ADC) with 20-nm gate length. Compared
with the Silicon FinFET (Si FinFET) ADC, the HTFET SAR ADC
achieves approximately 3 times power consumption reduction
and 6 times size reduction. Signal-to-noise and distortion ratio
is 31.4 dB for the HTFET SAR ADC, which is 2.81 dB higher
than the Si FinFET ADC due to the decreased quantization noise
rising from the high ON-current characteristic of HTFET at low
supply voltage. The energy per conversion step for both HTFET
and Si FinFET ADC designs are 0.43 and 1.65 fJ/conversion-step,
respectively, at a fixed supply voltage of 0.30 V.
Index Terms— Analog-to-digital converter (ADC), heterojunction tunnel FET (HTFET), successive-approximation register
(SAR), ultra-low-power.
U
I. I NTRODUCTION
LTRA-LOW-POWER circuit design techniques have
brought in growing interest in power-constrained
applications, such as energy harvesting systems, sensor
networks, and biomedical implants [1]–[3], where the energy
efficiency and area cost to convert analog signal to digital
data have profound impact on the overall system performance.
Tremendous progress has been made to leverage the power
consumption, chip area, and data conversion speed in analogto-digital-converter (ADC) designs to enable the low-power
mixed-signal/RF applications [2]–[5]. For digital circuits
in ADCs, technology scaling companied with the supply
voltage (VDD ) reduction provides continuous improvement of
energy efficiency [4]. However, the diminished signal-to-noise
ratio (SNR) at a lower VDD can be detrimental for analog
Manuscript received August 8, 2014; revised September 12, 2014; accepted
September 16, 2014. Date of current version October 20, 2014. This work
was supported in part by STARnet, a Semiconductor Research Corporation
Program, through MARCO and the Defense Advanced Research Projects
Agency, in part by the National Science Foundation through the Advanced
Self-Powered Systems of Integrated Sensors and Technologies Nanosystems,
European Research Council, under Award EEC-1160483, and in part by the
Intel Academic Research Office. The review of this paper was arranged by
Editor C. K. Sarkar.
M. S. Kim, X. Li, and V. Narayanan are with the Department of
Computer Science and Engineering, Pennsylvania State University, University
Park, PA 16802 USA (e-mail: [email protected]; [email protected];
[email protected]).
H. Liu and S. Datta are with the Department of Electrical Engineering,
Pennsylvania State University, University Park, PA 16802 USA (e-mail:
[email protected]; [email protected]).
Color versions of one or more of the figures in this paper are available
online at http://ieeexplore.ieee.org.
Digital Object Identifier 10.1109/TED.2014.2359663
Fig. 1. ADC data survey showing current technology limit and a noise limit
in current CMOS ADC designs [4], [7].
circuits [4], [5]. Recently, the near/subthreshold CMOS
technologies have been applied to ADC designs with digital
assisted blocks to explore the optimal energy efficiency albeit
with sacrificing certain degree of speed, matching, noise
performance, and area [4], [6]. However, the minimum energy
achieved in current ADC designs, as shown in Fig. 1 [4], is
still limited by the energy efficiency of CMOS technology,
especially in the low-resolution [low signal-to-noise and
distortion ratio (SNDR)] regime, where the innovations of
device technology are required to enable further energy
reduction beyond CMOS limit [4], [7].
Tunnel field-effect transistor (TFET) [8] has emerged as
a strong candidate for various low voltage digital applications [9]–[11] due to its superior energy efficiency arising
from the subthermal switching characteristics at a low VDD .
The work on TFET analog/RF circuit designs in [12]
and [13] further explored its unique device characteristics,
such as asymmetrical source/drain design, steep switching,
and low voltage operation, to construct the ultra-low
power energy harvesting systems. The recent work in [14]
demonstrated the high-frequency switching of the fabricated In0.9 Ga0.1 As/GaAs0.18 Sb0.82 near broken-gap heterojunction tunnel FET (III–V HTFET) with intrinsic cutoff
frequency ( f T ) of 22 and 39 GHz achieved at 0.30 and 0.50 V,
respectively, which highlighted the desired RF characteristics and significant power reduction projection with channellength scaling. Therefore, TFET technology can be potentially
applied to overcome the growing challenge in energy
efficiency using the conventional CMOS technology for ultralow power mixed-signal applications.
The goal of this paper is to explore the advantages of 20-nm
HTFET device characteristics in mixed-signal domain and to
0018-9383 © 2014 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
3662
IEEE TRANSACTIONS ON ELECTRON DEVICES, VOL. 61, NO. 11, NOVEMBER 2014
design energy efficient HTFET ADC operating below 0.50 V.
For the first time, we explore the design, analysis and performance benchmarking of the III–V HTFET 6-bit successiveapproximation register (SAR) ADC (HTFET SAR ADC)
against the Si FinFET SAR ADC for ultra-low power/energyefficient systems. Among various ADC designs, we choose the
SAR ADC topology due to its desired energy efficiency in low
to moderate resolution regime and medium bandwidth application ranges [4], [15], where low-power devices are desired for
energy saving purpose (Fig. 1). We also investigate the device
noise impact on the HTFET SAR ADC performance compared
with the Si FinFET SAR ADC. This paper is organized as
follows. Section II presents the advantage of the HTFET
for ultra-low power SAR ADC and the device modeling
methodology for the HTFET circuit implementations.
The design of individual modules of both HTFET and
Si FinFET SAR ADCs are described in Section III.
Section IV presents the energy and performance evaluation
for both HTFET and Si FinFET SAR ADCs. The conclusion
is provided in Section V to summarize our results.
II. TFET A DVANTAGES AND M ODELING M ETHODOLOGY
FOR L OW-P OWER ADC A PPLICATION
A. Low-Power CMOS ADC Design Challenges
In digital circuits, the dynamic power consumption PDigital is
quadratic related to VDD , as shown in (1), hence, VDD scaling
can enable the power reduction with the effective control of
the OFF-state leakage power [5]. Unlike the digital circuits,
the analog circuit performance is primarily limited by the
thermal noise kT/C (k is the Boltzmann constant, T is the
absolute temperature, and C is the overall capacitance), which
is inversely proportional to the current IDS at a given bias
point (kT/C ∝ 1/IDS ) [4], [5]. The power dissipation in the
SNR limited analog circuits can be expressed as (2) assuming
the signal power is (β · VDD )2 [4], [5]
2
× α × fs
PDigital ∝ C × VDD
1
1
1
× gm × kT × SNR × fs .
PAnalog ∝ ×
β
VDD
IDS
(1)
(2)
Assuming
(βV DD )2
gm
, SNR ∝
C
kT /C
where α is the activity factor, which indicates the ratio of the
internal clock frequency f c over sampling frequency f s , β is
the ratio of signal peak voltage Vpp to VDD , gm is the transistor
transconductance, and the transistor bias point is indicated
by gm /IDS . Therefore, in a fixed design with constant gm /IDS
and f s diminishing VDD, while maintaining SNR will worsen
the power consumption in analog components.
Today’s analog system designs normally employ digital
components to assist the performance and functionality
[4], [5], e.g., SAR ADCs. The total power consumption can
be expressed as PADC = PAnalog + PDigital, which leads to the
total energy consumption E ADC as (at β = 1)
!
"
#
$
P
1
1
2
∝
× gm ×kT ×SNR + C ×VDD
×α . (3)
E ADC =
fs
VDD IDS
fs ∝
Fig. 2. (a) 2-D TCAD simulation schematic of the double-gate GaSb-InAs
HTFET [10]. (b) IDS –VGS characteristics comparing Si FinFET and
HTFET at L g = 20 nm obtained from calibrated TCAD models in
[10] and [16], respectively. (c) n-HTFET device parameters for TCAD simulation. For HTFET, 1-nm gate-source overlap and 1-nm gate-drain underlap
are used in the model.
In CMOS technology, gm /IDS has the maximum value
of 40 V−1 achieved in the subthreshold region as
!
!
""
!
"
I D0
VDS
VGS−Vth
gm,MOSFET =
1−exp −
· exp
(4)
nVt
Vt
nVt
1
1
gm,MOSFET
≈ 40 V−1
∝
<
(5)
IDS
nVt
26 mV
where Vth is the threshold voltage and Vt is the thermal
voltage. However, subthreshold operation of CMOS can significantly reduce the transistor f T , which limits the sampling
frequency f s due to the requirement of f s < f T /80 [4].
Fig. 1 shows the energy consumption of various ADC designs
versus SNDR (with the consideration of signal-distortion
effect in ADCs) with the fundamental noise limit given by the
minimum energy of the idealized amplifier E min and practical
amplifier E class_A [4], which reveals the energy efficiency
challenge in CMOS ADC design. The high-resolution design
(SNDR > 60 dB) is limited by the noise-limited power
efficiency and nonlinearity matching, while the low-resolution
design (SNDR < 60 dB) is limited by state-of-the-art CMOS
technology. The limited gm /IDS and practical tradeoff of
gm /IDS versus f T pose the fundamental challenge for VDD
reduction in ultra-low power CMOS ADCs.
B. TFET Advantages for Low-Power ADC Design
TFET overcomes the 60 mV/dec subthreshold slope (SS)
limit of MOSFET [8] due to the tunneling induced carrier
injection mechanism. With improved tunneling probability
and high ON-current at a low VDD achieved by the III–V
HTFET [11], [17], VDD scaling can be further enabled to
mitigate the challenge between the leakage power constrain
and Vth scaling in the TFET digital circuits. Here, we setup
the GaSb-InAs HTFET model using technology computeraided design (TCAD) Sentaurus [16], which has been calibrated [10] with full-band atomistic simulation in [18]. The
Si FinFET is used as baseline for performance comparison,
which has also been calibrated with experimental data in [19].
As shown in Fig. 2, an effective SS of 30 mV/dec (over two
KIM et al.: STEEP-SLOPE TUNNEL FET BASED SAR ADC
3663
Fig. 4. III–V p-HTFET simulation structure and device parameters for circuit
simulations.
TABLE I
D EVICE C HARACTERISTICS C OMPARISON
Fig. 3. (a) gm /IDS versus VGS of HTFET showing HTFET overcoming
40 V−1 CMOS limit due to steep switching at a low VGS . (b) f T versus
gm /IDS comparison of HTFET and Si FinFET, showing improved design
space of HTFET considering the tradeoff between high frequency and gm /IDS .
decades change of IDS from 10−2 to 1 µA/µm) and 7 times
improvement of IDS at 0.30 V can be achieved in the modeled
HTFET over Si FinFET at gate-length L g = 20 nm, which is
consistent with the reported simulation results in [11]. The
device parameters for our modeling are listed in Fig. 2(c) for
the HTFET and the baseline Si FinFET n-type devices.
Fig. 3(a) shows the gm /IDS versus VGS for HTFET and
Si FinFET at VDS = 0.30 and 0.50 V, respectively. The
steep slope of the HTFET provides significant improvement
of gm /IDS to overcome the CMOS limit (40 V−1 ) given
gm,HTFET
∂ IDS 1
∂lnI DS
ln10
> 40 V−1 . (6)
=
=
=
IDS
∂ VGS IDS
∂ VGS
SS
As discussed in Section II-A, high gm /IDS of HTFET
can mitigate the power increase with VDD scaling in analog
components. Moreover, the tradeoff between gm /IDS (40 V−1
achieved in subthreshold regime) and f T (peak f T achieved in
super-threshold regime) of CMOS [4], [5] can be further eliminated in HTFET due to the high gm /IDS (energy efficiency) and
desired f T ( f s requirement) can be achieved simultaneously.
Fig. 3(b) shows the f T versus gm /IDS comparing between
HTFET and Si FinFET, where f T is calculated from
gm /2π(Cgs + Cgd ). Cgs and Cgd are the gate-source and the
gate-drain capacitances, respectively. Therefore, the HTFET
can provide energy saving in both digital and analog components for low VDD in terms of circuit speed, drive strength, and
improved design space between operation point and bandwidth
requirement, which is highly desirable for ultra-low power
ADC design.
C. Device-Circuit Modeling Methodology
P-type TFET is required for complementary circuit design.
However, the challenge in realizing III–V p-type HTFET
(p-HTFET) still remains due to the low density of states
in the conduction band of III–Vs, which leads to a large
portion of the temperature dependent part of the source
Fermi function participating in tunneling and a temperature
dependent SS [20]. Here, we model a complementary III–V
p-HTFET with symmetrical material system as III–V n-HTET,
as shown in Fig. 4. The modeled p-HTFET shows degraded SS
Fig. 5. Lookup table based Verilog-A model schematic for double-gate
HTFET and Si FinFET. Both dc and transient characteristics can be captured.
of 55 mV/dec over 2 decades of current change comparing
with the n-HTFET at VDD = 0.30 V (Table I).
To implement the HTFET based circuits, we have developed
device-circuit co-design framework based on the Verilog-A
models in Fig. 5 [9], [10], [12], [13], [21]. The HTFET
Verilog-A model is built from 2-D lookup tables including
IDS (VGS , VDS ), Cgs (VGS , VDS ), and Cgd (VGS , VDS ), which
are obtained from fine-granularity simulation across a wide
range of VDS and VGS using TCAD Sentaurus. Considering
the circuit complexity and number of transistors (over 1100)
in an ADC simulation, we omitted the drain–source capacitances Csd given its relatively small value compared with
Cgs and Cgd [28]. This simplification allows us to simulate
the ADC within the simulation capability of the lookup tablebased model [29]. In addition, the source and drain terminal
charges, Q S and Q D , are computed with respect to the
terminal voltages and capacitances, respectively, in transient
analysis that preserves the charge conservation. Therefore,
both dc and transient characteristics can be captured using
this modeling method, which has also been applied by
other works in [13], [23], and [24]. Furthermore, the recent
work in [25] has reported the transient characteristics of
the fabricated Si TFET inverter, showing enhanced ON-state
3664
IEEE TRANSACTIONS ON ELECTRON DEVICES, VOL. 61, NO. 11, NOVEMBER 2014
Fig. 7.
Sizing strategies of the logic gates. (a) HTFET. (b) Si FinFET.
Fig. 6. (a) Principle block of 6-bit SAR ADC. (b) SAR ADC design transistor
sizing: total transistor count versus device width and sizing strategy for critical
analog and digital blocks.
Miller capacitance effect in TFET agreed with [26] from
TCAD simulation. Bijesh et al. [14] have also confirmed the
feasibility of the HTFET TCAD model for high-frequency
performance analysis, showing good agreement of extracted
capacitance values and f T between simulation results and RF
measurements.
Note that the frequency of the input signal, f in , in our
ADC design is between 50 and 400 KHz, and a sampling
frequency, f s , of 10 MS/s, which are significantly below
the f T of the transistor. Hence, the quasi-static assumption
is fulfilled for our modeling technique.
To analyze the device noise effect on the SAR ADC
performance (e.g., SNDR and power), we use the noise
models developed in [21] for the HTFET and the baseline
Si FinFET, including thermal, shot, and flicker noise. Table I
shows the key device metrics for the ADC designs. We use
the effective Vth of HTFET from the linear extrapolation
of IDS –VGS for and overdrive voltage and transistor sizing
estimation in the analog component. In the following
sections, we implement the HTFET SAR ADC and perform
the performance evaluation based on this Verilog-A model.
We vary the transistor width W , while keeping L g constant
(20 nm) for transistor sizing.
III. HTFET 6-B IT SAR ADC D ESIGN
Circuit design using HTFET requires certain modifications
due to the change in the device architecture and characteristics
[9], [10], [12]–[14]. The asymmetrical source/drain of the
HTFET results in the unidirectional conduction characteristic, which requires the modification of pass transistor logic,
latches, flip-flops, and so on [9], [10]. The low effective Vth
due to the steep SS of HTFET also requires redesign of
the comparator reference voltage and bias current circuitry to
fulfill the timing difference. In this section, we will present
the detailed design modification, sizing strategies, and performance analysis of each digital and analog block for the HTFET
6-bit SAR ADC.
A. Circuit Implementation of HTFET SAR ADC
Fig. 6(a) shows the principal blocks of HTFET 6-bit
SAR ADC including the sample-and-hold (S/H) circuit,
comparator, self-determined SAR control logic, and switched
capacitor array operating as the feedback digital-to-analog
converter (DAC).
Fig. 8. (a) HTFET and Si FinFET comparator schematic. (b) Delay with
respect to input difference (LSB). (c) Cgd variations according to input
difference (LSB). (d) Power consumption versus input difference (LSB).
The key challenge of designing SAR ADC using Si FinFET
at a low VDD is the accurate operation of the analog blocks
(comparator and DAC) with the insufficient drive current at
near/subthreshold. In addition, to meet the timing requirement
in digital blocks and to compensate the voltage headroom
reduction due to a low VDD [4], [5], the new size strategy for
Si FinFET has been adopted to use a smaller fan-out factor
between logic gates, as shown in Fig. 7. To drive the same
analog load (switches), the Si FinFET logic gates have an
average size of around 6–7 times larger than the HTFET gates
with given the device characteristics in Fig. 2(b). Some sizing
adjustment in certain transistors is applied to realize the circuit
functionality.
Circuit-level implementation and performance analysis are
performed using Cadence Spectre [27] based on the Verilog-A
modeling technique in Section II-C. The device models include
the electrical noise (flicker, shot, and thermal) for both HTFET
and Si FinFET. To explore the HTFET benefits at a low VDD ,
as described in Section II-B, we focus on VDD = 0.30, 0.40,
and 0.50 V for circuit implementation at fixed f s = 10 MS/s.
B. HTFET Comparator Design
A two-stage dynamic comparator in Fig. 7(a) is used due
to its zero static biasing with double-gain stage to operate at a
low VDD [3]. Due to the steep slope characteristic of HTFET
at a low VDD , the simulated input offset voltage of the HTFET
comparator is only 0.9 mV, which is much less than 9.5 mV
in the Si FinFET comparator.
Fig. 8(b) shows two orders of magnitude reduction in delay
and less delay variation over a wide range of input voltage are
KIM et al.: STEEP-SLOPE TUNNEL FET BASED SAR ADC
Fig. 9.
Two types of DFF schematics; standard and new design schematic.
observed in the HTFET comparator due to faster node voltage
transition that reduces the nonlinearity and the quantization
error in ADC. With comparable OFF-state leakage for the
Si FinFET and HTFET. The Si FinFET comparator consumes
higher dynamic power due to the increased capacitance from
the up-sizing. Even though the enhanced ON-state Miller
capacitance effect of HTFET [26] in Fig. 8(b) causes slightly
dynamic power increase overall 3.5 times power reduction
is achieved in the HTFET comparator over the Si FinFET
comparator.
C. Feedback DAC
The feedback DAC is implemented with a binary-scaled
charge-distribution topology [28]. Digital bits (D0 –D5 ) from
SAR drive the bottom of capacitors to generate the output
of DAC. An overall capacitance 76.6 fF is used in our design
to meet the thermal noise (kT/C) limitation [3], [4], [30], [31].
D. HTFET SAR Logic
Fig. 6(a) shows the SAR logic block as the main digital
block, which generates the relevant control bits (D0 –D5 )
from tracing the output of comparator. To avoid using
an external frequency generator, SAR is constructed with
D flip-flops (DFFs), and the clock-manage logic is utilized
to set the enable signal from/to the comparator to generate
internal clock signals [28]. Binary decoder in Fig. 6(a) is
used to convert binary-scaled control bits into binary codes.
It accepts control bits and an internal synchronous clock signal
f c (from SAR logic to generate the binary outputs).
Since DFF is the main component of digital blocks in SAR
logic, the power reduction of the HTFET DFF dominates the
overall power reduction of ADC. Two types of DFF designs
are evaluated for energy efficiency optimization for SAR logic,
as shown in Fig. 9. The HTFET logic gate DFF (Type 1)
requires no design change, while the HTFET transmission
gate DFF (Type 2) requires additional transistors for charging/
discharging of internal nodes due to its unidirectional
conduction. Still, the HTFET DFF outperforms the Si FinFET
DFF with lower power and delay. Between two types of
DFF designs, type 1 DFF was utilized in this ADC design
because type 1 DFF exhibits balanced performance, as shown
in Fig. 10, in terms of delay and power consumption. Due to
the 20 times current improvement of HTFET over Si FinFET
at the low VDD limit, both analog and digital blocks show
3665
Fig. 10. (a) Delay and (b) power consumption comparison between two
types of both HTFET and Si FinFET DFFs.
Fig. 11. ADC analog input. DAC output waveform analysis and output
digits evaluation comparison between HTFET and Si FinFET ADCs
at VDD = 0.30 V.
the average 6 times reduction in transistor size and 3 times
reduction in power consumption.
IV. E NERGY AND P OWER E VALUATION
The single-ended SAR ADC design accepts an analog input
ranging from 0 V to VDD as a full-scaled input voltage. Fig. 11
shows the input waveform and digitized DAC output of both
HTFET and Si FinFET ADCs at 0.30 V, where the HTFET
ADC shows clear transitions of voltage steps as compared
with Si FinFET ADC. An improved effective number of
bits (ENOB) and SNDR with respect to the analog input
frequency and supply voltage are achieved in the HTFET
ADC, as shown in Fig. 12, due to the improved device drive
strength at a low VDD as compared with the Si FinFET ADC
design. A continuous degradation with respect to the input
frequency is observed for the Si FinFET design, due to the
increased quantization errors with the increased nonlinearity,
and the existence of electrical noise. Since flicker noise is
dominant in megahertz frequency range (where white noise
can be neglected), the reduced flicker noise in HTFET [21]
leads to a slower degradation of ENOB and SNDR at lower
input frequency than the Si FinFET ADC.
Energy per conversion-step has been widely used as figure
of merit (FoM) of ADCs [3], [6], [28], [30]. As shown
in Fig. 12(d), a FoM of 0.43 fJ/conversion-step is achieved
in the HTFET ADC at 0.30 V, which is 3.8 times lower than
1.65 fJ/conversion-step of the Si FinFET ADC. Fig. 13 shows
the differential nonlinearity (DNL) and the integral nonlinearity (INL) performance. The HTFET ADC shows the maximum
+0.90/−2.1 LSB (DNL) and +1.0/−2.5 LSB (INL), which
are 35.2% (DNL) and 5.2% (INL) lower compared with the
Si FinFET ADC. This comes from the improved bit decisions
3666
IEEE TRANSACTIONS ON ELECTRON DEVICES, VOL. 61, NO. 11, NOVEMBER 2014
Fig. 14. HTFET and the Si FinFET SAR ADCs power analysis for analog
and digital blocks at VDD = 0.30 V. Digital blocks consume major power for
the HTFET and Si FinFET ADCs. The HTFET ADC shows up to 3 times
power reduction for each block comparing with Si FinFET ADC.
Fig. 12. Comparison of (a) ENOB and (b) SNDR versus input frequency.
(c) ENOB and (d) FoM versus supply voltage for both HTFET and Si FinFET
ADCs. The HTFET ADC shows higher ENOB and SNDR compared with
Si FinFET ADC.
Fig. 15. Zoomed-in view of Fig. 1. Both HTFET and Si FinFET 6-bit SAR
ADCs’ energy versus SNDR evaluation for VDD = 0.30, 0.40, and 0.50 V.
Fig. 13.
(a) DNL and (b) INL analysis comparisons.
TABLE II
P ERFORMANCE M ETRICS OF HTEFT AND Si FinFET SAR ADCs
28.6 and 33.5 dB of SNDR, respectively, for the Si FinFET
ADC design. The HTFET ADC shows remarkable reduction in
energy ranging from 3.19 × 106 to 11.0 × 106 kT with higher
a SNDR range of 31.4–34.6 dB, resulting from resizing of
devices and higher gm /IDS at a low VDD . The HTFET ADC
design offers promise for further energy reduction capability
and SNDR improvement toward the noise limit with improved
energy efficiency.
V. C ONCLUSION
of the HTFET comparator at very low input offset, as well
as improved linearity due to the high drive current strength
of HTFET. It is consistent with Fig. 11, where the output of
DAC of HTFET shows reduced nonlinearity during sampling
and amplification [1], [3], [30]. The static nonlinearity can
be further improved with the calibration block and differential
autozeroing DAC topology for the reduction of offset. The performance metrics are summarized in Table II at different VDD .
Power analysis (Fig. 14) shows 3 times power reduction
in majority of the digital and analog blocks in the HTFET
over the Si FinFET ADCs, with SAR logic block being the
most power hungry component. Energy (P/ f s ) metrics for
HTFET and Si FinFET ADCs are presented in Fig. 15. The
energy for both ADC designs is lower than current technology
boundary due to scaling and device performance improvement
(for the Si FinFET and the HTFET). The energy consumption
is between 8.84 × 106 and 49.0 × 106 kT in the range of
Steep-slope HTFET based energy efficient 6-bit SAR ADC
has been evaluated from the device to the circuit level. HTFET
ADC shows significant improvement in energy efficiency compared with the Si FinFET ADC design baseline. We demonstrate the HTFET SAR ADC with 2.81 dB higher SNDR,
0.47-bit larger ENOB, 63% lower average power consumption and 26% lower energy per conversion-step (FoM). This
performance advantage stems from increasing current strength
due to steep-slope transistor characteristics and transistor size
reduction, which is desirable for ultra-low power ADCs at
VDD < 0.50 V.
R EFERENCES
[1] R. Sarpeshkar, Ultra Low Power Bioelectronics: Fundamentals, Biomedical Applications, and Bio-Inspired Systems. Cambridge, U.K.:
Cambridge Univ. Press, 2010.
[2] M. D. Scott, B. E. Boser, and K. S. J. Pister, “An ultralow-energy
ADC for smart dust,” IEEE J. Solid-State Circuits, vol. 38, no. 7,
pp. 1123–1129, Jul. 2003.
[3] M. van Elzakker, E. van Tuijl, P. Geraedts, D. Schinkel,
E. A. M. Klumperink, and B. Nauta, “A 10-bit charge-redistribution
ADC consuming 1.9 µW at 1 MS/s,” IEEE J. Solid-State Circuits,
vol. 45, no. 5, pp. 1007–1015, May 2010.
KIM et al.: STEEP-SLOPE TUNNEL FET BASED SAR ADC
[4] B. Murmann, “A/D converter trends: Power dissipation, scaling and
digitally assisted architectures,” in Proc. IEEE Custom Integr. Circuits
Conf. (CICC), Sep. 2008, pp. 105–112.
[5] A. Baschirotto, V. Chironi, G. Cocciolo, S. D’Amico, M. De Matteis, and
P. Delizia, “Low power analog design in scaled technologies,” in Proc.
Topical Workshop Electron. Particle Phys., Paris, France, Sep. 2009,
pp. 103–110.
[6] C.-Y. Chen, J. Wu, J.-J. Hung, T. Li, W. Liu, and W.-T. Shih, “A 12-bit 3
GS/s pipeline ADC with 0.4 mm2 and 500 mW in 40 nm digital CMOS,”
IEEE J. Solid-State Circuits, vol. 47, no. 4, pp. 1013–1021, Apr. 2012.
[7] B. Murmann. (Jul. 2013). ADC Performance Survey 1997-2013. [Online]
Available: http://http://web.stanford.edu/∼murmann/adcsurvey.html
[8] A. C. Seabaugh and Q. Zhang, “Low-voltage tunnel transistors for
beyond CMOS logic,” Proc. IEEE, vol. 98, no. 12, pp. 2095–2110,
Dec. 2010.
[9] V. Saripalli, G. Sun, A. Mishra, Y. Xie, S. Datta, and V. Narayanan,
“Exploiting heterogeneity for energy efficiency in chip multiprocessors,”
IEEE J. Emerg. Sel. Topics Circuits Syst., vol. 1, no. 2, pp. 109–119,
Jun. 2011.
[10] V. Saripalli, S. Datta, V. Narayanan, and J. P. Kulkarni, “Variationtolerant ultra low-power heterojunction tunnel FET SRAM design,”
in Proc. IEEE/ACM Int. Symp. Nanosc. Archit., vol. 1. Jun. 2011,
pp. 45–52.
[11] U. E. Avci, S. Hasan, D. E. Nikonov, R. Rios, K. Kuhn, and I. A. Young,
“Understanding the feasibility of scaled III–V TFET for logic by
bridging atomistic simulations and experimental results,” in Proc. Symp.
VLSI Technol. (VLSIT), Jun. 2012, pp. 183–184.
[12] H. Liu, R. Vaddi, S. Datta, and V. Narayanan, “Tunnel FETbased ultra-low power, high-sensitivity UHF RFID rectifier,” in Proc.
IEEE Int. Symp. Low Power Electron. Design (ISLPED), Sep. 2013,
pp. 157–162.
[13] A. R. Trivedi, S. Carlo, and S. Mukhopadhyay, “Exploring tunnel-FET
for ultra low power analog applications: A case study on operational
transconductance amplifier,” in Proc. 50th ACM/EDAC/IEEE Design
Autom. Conf. (DAC), May/Jun. 2013, pp. 1–6.
[14] R. Bijesh et al., “Demonstration of In0.9 Ga0.1 As/GaAs0.18 Sb0.82
near broken-gap tunnel FET with ION=740 µA/µm, GM=70 µS/µm
and gigahertz switching performance at VDs=0.5V,” in Proc. IEDM,
Dec. 2013, pp. 28.2.1–28.2.4.
[15] P. J. A. Harpe et al., “A 26 µW 8 bit 10 MS/s asynchronous SAR ADC
for low energy radios,” IEEE J. Solid-State Circuits, vol. 46, no. 7,
pp. 1585–1595, Jul. 2011.
[16] TCAD Sentaurus Device Manual, Synopsys, Mountain View, CA, USA,
2010.
[17] D. K. Mohata et al., “Demonstration of improved heteroepitaxy, scaled
gate stack and reduced interface states enabling heterojunction tunnel
FETs with high drive current and high on-off ratio,” in Proc. IEEE
Symp. VLSI Technol. (VLSIT), Jun. 2012, pp. 53–54.
[18] M. Luisier and G. Klimeck, “Performance comparisons of tunneling
field-effect transistors made of InSb, carbon, and GaSb-InAs broken gap
heterostructures,” in Proc. IEEE Int. Electron Devices Meeting (IEDM),
Dec. 2009, pp. 1–4.
[19] A. Nidhi, V. Saripalli, V. Narayanan, Y. Kimura, R. Arghavani, and
S. Datta, “Will strong quantum confinement effect limit low VCC
logic application of III–V FINFETs?” in Proc. 70th Annu. Device Res.
Conf. (DRC), Jun. 2012, pp. 231–232.
[20] S. Datta, R. Bijesh, H. Liu, D. Mohata, and V. Narayanan, “Tunnel
transistors for energy efficient computing,” in Proc. IEEE Int. Rel. Phys.
Symp. (IRPS), Monterey, CA, USA, Apr. 2013, pp. 6A.3.1–6A.3.7.
[21] R. Pandey, B. Rajamohanan, H. Liu, V. Narayanan, and S. Datta,
“Electrical noise in heterojunction interband tunnel FETs,” IEEE Trans.
Electron Devices, vol. 61, no. 2, pp. 552–560, Feb. 2014.
[22] L. Zhang, X. Lin, J. He, and M. Chan, “An analytical charge model
for double-gate tunnel FETs,” IEEE Trans. Electron Devices, vol. 59,
no. 12, pp. 3217–3223, Dec. 2012.
[23] Predictive Technology Model (PTM). (Feb. 2008). Post-Silicon Devices.
[Online] Available: http://ptm.asu.edu/cnt-fet/
[24] U. E. Avci et al., “Energy efficiency comparison of nanowire heterojunction TFET and Si MOSFET at Lg=13 nm, including P-TFET and
variation considerations,” in Proc. IEDM, Dec. 2013, pp. 33.4.1–33.4.4.
[25] L. Knoll et al., “Inverters with strained Si nanowire complementary
tunnel field-effect transistors,” IEEE Trans. Electron Devices, vol. 34,
no. 6, pp. 813–815, Jun. 2013.
[26] S. Mookerjea, R. Krishnan, S. Datta, and V. Narayanan, “On enhanced
Miller capacitance effect in interband tunnel transistors,” IEEE Electron
Device Lett., vol. 30, no. 10, pp. 1102–1104, Oct. 2009.
[27] Virtuoso Spectre Circuit Simulator, Cadence, San Jose, CA, USA, 2009.
3667
[28] C.-C. Liu, S.-J. Chang, G.-Y. Huang, and Y.-Z. Lin, “A 10-bit 50-MS/s
SAR ADC with a monotonic capacitor switching procedure,” IEEE J.
Solid-State Circuits, vol. 45, no. 4, pp. 731–740, Apr. 2010.
[29] H. Liu, V. Saripalli, V. Narayanan, and S. Datta. (May 2014).
III-V
Tunnel
FET
Model
1.0.0.
[Online].
Available:
http://nanohub.org/resources/21012
[30] J. Hu, N. Dolev, and B. Murmann, “A 9.4-bit, 50-MS/s, 1.44-mW
pipelined ADC using dynamic residue amplification,” in Proc. IEEE
Symp. VLSI Circuits, Jun. 2008, pp. 216–217.
[31] M. Furuta, M. Nozawa, and T. Itakura, “A 10-bit, 40-MS/s, 1.21 mW
pipelined SAR ADC using single-ended 1.5-bit/cycle conversion technique,” IEEE J. Solid-State Circuits, vol. 46, no. 6, pp. 1360–1370,
Jun. 2011.
Moon Seok Kim (S’13) received the B.S. degree
in electronic engineering from Inha University,
Incheon, Korea, in 2009, and the M.S. degree in
electrical engineering from Northeastern University,
Boston, MA, USA, in 2012. He is currently pursuing
the Ph.D. degree with the Department of Computer
Science and Engineering, Pennsylvania State University, University Park, PA, USA.
Huichu Liu (S’11) received the B.S. degree in
microelectronics from Peking University, Beijing,
China, in 2009. She is currently pursuing the Ph.D.
degree with the Department of Electrical Engineering, Pennsylvania State University, University Park,
PA, USA.
She has been a Research Assistant with Pennsylvania State University since 2009, and with the
Nanoelectronic Devices and Circuits Laboratory and
the Microsystem Design Laboratory since 2011.
Xueqing Li (M’14) received the B.S. and Ph.D.
degrees in electronics engineering from Tsinghua
University, Beijing, China, in 2007 and 2013,
respectively.
He has been a Post-Doctoral Researcher with the
Department of Computer Science and Engineering,
Pennsylvania State University, University Park, PA,
USA, since 2013.
Suman Datta (F’13) received the B.S. degree in
electrical engineering from IIT Kanpur, Kanpur,
India, in 1995, and the Ph.D. degree in electrical
and computer engineering from the University of
Cincinnati, Cincinnati, OH, USA, in 1999.
He was the Joseph Monkowski Associate
Professor
of
Electrical
Engineering
with
Pennsylvania State University, University Park, PA,
USA, in 2007, where he is currently a Professor of
Electrical Engineering.
Vijaykrishnan Narayanan (F’11) received the B.S.
degree in computer science and engineering from the
University of Madras, Chennai, India, in 1993, and
the Ph.D. degree in computer science and engineering from the University of South Florida, Tampa,
FL, USA, in 1998.
He is currently a Professor of Computer Science
and Engineering and Electrical Engineering with
Pennsylvania State University, University Park, PA,
USA.